|Publication number||US7016991 B2|
|Application number||US 10/801,686|
|Publication date||Mar 21, 2006|
|Filing date||Mar 17, 2004|
|Priority date||Apr 9, 2001|
|Also published as||US6779063, US6944690, US7035950, US7167960, US7174399, US7404053, US7788459, US20020152339, US20040133718, US20040177174, US20040177181, US20040243732, US20060149868, US20070088880, US20080270688|
|Publication number||10801686, 801686, US 7016991 B2, US 7016991B2, US-B2-7016991, US7016991 B2, US7016991B2|
|Original Assignee||Hitachi, Ltd.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (31), Referenced by (2), Classifications (39), Legal Events (2)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The present application is a continuation of application Ser. No. 09/829,470, filed Apr. 9, 2001, now U.S. Pat. No. 6,779,063, the contents of which are incorporated herein by reference.
The present invention relates generally to data processing systems, and particularly to a direct access storage system with a combined block interface and file interface access.
Interconnecting the various elements of a processing system (e.g., processor units and peripheral equipment such as direct access storage devices) permits the resources of the system to be distributed so that they are available to all elements of the system. For example, multiple processor units may be connected to a storage system for sharing not only the afforded storage space, but the files that are stored there. Typically, a network architecture of one type or another will be used to implement the interconnection, which may dictate the particular of interface structure between the elements of a system, e.g., a processor unit and a data storage system. For example, it has been popular to connect stand-alone processor units to direct access storage devices using a small computer standard interface (SCSI). SCSI connections use block transfer protocols in which a logical unit number (LUN) identifies the logical volume for access.
Network protocols, on the other hand, are different. Protocols of choice for networked and distributed processing systems included Network File System (“NFS;” an open operating system developed by Sun Microsystems), a Common Internet File System protocol (“CIFS;” a remote file access protocol), or a HyperText Transport Protocol, more popularly known as “HTTP.” These protocols use what is known as a “file system interface,” and while the file interface structures used to implement the different file system interface protocols, they use a common file system structure. Thus, data stored on a storage system using a file system interface of two or more types are available to all host systems. For example, a storage system capable of handling input/output requests of both NFS and CIFS protocols, i.e., an NFS protocol interface and a CIFS protocol interface, can store data files that are accessible to host processors having either of the NFS interfaces. That is, a host system with only an NFS interface can access and open files stored by a host system with a CIFS interface, and the host system with a CIFS interface can access and open files stored by the system via the NFS interface—provided the storage system has both interfaces.
Storage systems having one or more of the file system interfaces of the types described above provide access through an I/O read or write request that includes a file name, and an lock request that seeks a right to access the particular file of the I/O request.
Most direct access storage systems have either a block interface or a file interface, and host systems using a block interface protocol cannot access storage systems employing file interface protocols. Further, because of the differences between block and file interface structures and the way data is stored and accessed, a storage system is structured for a block system or a file system, but not both.
The present invention provides a storage system with direct access storage devices that can be shared between a block interface and a file interface. The invention provides a system architecture with both block and file interfaces to realize high performance, scalability, and availability.
According to the present invention a storage system includes a plurality of physical disk units, a host processing system that may include a number of processing units, and a controller element that includes a SCSI interface adapted to receive block type read/write requests and at least one file system interface adapted to receive I/O read/write file requests. The file system interface may be compatible with a network file system (NFS), a Common Internet File System (CIFS) protocol or HyperText Transfer Protocol (HTTP), or any combination of file system protocols. The controller element operates to connect the processor units of the host processing system to the plurality of physical disk units. The controller unit uses logical volume management, allowing the different block and file system I/O requests to access portions of the physical disk units allocated for block system data or file system data.
In an alternate embodiment of the invention, file system data stored on the physical disk units is made accessible to a block system request by performing a volume backup, thereby permitting data sharing between a SCSI interface and a file system interface. A number of advantages are achieved by the present invention. First, is that direct access storage device (“DASD”) resources can be shared between those processing elements having only a block interface, and those processing elements having a file system interface or multiple file system interfaces.
Turning now to the figures, and first to
Although not specifically shown, the host system 12 most likely will comprise a plurality of processor units, although it could also comprise a single processor unit with multiple I/O interfaces, including a block system interface and at least one file system interface. It should be understood, therefore, that the host system however implemented will include at least one SCSI protocol type interface (for block system file transfers with the storage controller 14) and at least one file system interface, such as an interface or interfaces that operate according to NFS, CIFS, and/or HTTP protocols. Accordingly, the host system may comprise multiple processor units, one having an SCSI interface, another with an NFS interface, still another with a CIFS interface, and so on. Alternatively, the host system may be implemented by a single processor unit having all four (SCSI, NFS, CIFS, and HTTP) type interfaces.
The host system 12 is coupled to the storage controller 14 by a bus structure 16. For reasons that will become clearer below, the bus system 16 may be multiple bus structures to connect the host system to corresponding ones of four interface adaptors 26-32 of the storage controller 14.
The storage system 14 may have any number of any type of the interface adapters 26-32. For example, a storage controller 14 configuration may include two (2) SCSI interface adaptors 26, one (1) NFS interface adaptor 28, three (3) CIFS interface adaptors 30, and two (2) HTTP interface adaptors 32. Alternatively, another storage controller 14 configuration may have just four interface adapters, one of each type, with the capability of having more adapters of any type added. As can be seen, a variety of other alternative storage controller configurations are possible. By providing the storage controller 14 with such a flexible architecture, high scalable performance and high availability is achieved. This, in turn, provides a storage system controller 14 with the capability of increasing, for example, the number of NFS interface type adapters according to performance demands placed upon the storage system by the host system 12. Moreover, by providing the storage controller 14 with multiple interface adapters of the same type (e.g., NFS interface adapters) a failure of one still leaves the other or others of that same type to execute the requested processing from the host system.
The storage controller 14 also includes a terminal interface adaptor 43 to provide a system administrator with access to the storage controller for configuration purposes, as will be discussed more fully below.
Referring now to
As will be seen, the storage controller 14 employ a logical volume management in order to share the resources of the physical disk units 20 between block system and file system interfaces. Accordingly, the SCSI interface adaptor includes a logical volume access block function 64 that is configured to convert the LUN of a I/O read write request to a logical volume access. Also included in the SCSI interface adapter 26 is a drive interface adaptor (DIA) interface function 66 to handle the communication colloquy with the drive interface adaptors 46 in response to information provided by the logical volume access block 64. A conventional cache manager function 68 manages data access of the SCSI interface adapter 26 to the cache memory 42 by the SCSI interface adaptor 26.
The NFS interface adaptor 28 is functionally illustrated in FIG. 3. The other file system interface adapters, i.e., the CIFS and HTTP interface adapters are functionally equivalent to the NFS interface adapter, with the exception of the process block 72, so that the description of the NFS interface adapter 28 will apply equally to the CIFS and HTTP interface adapters 30 and 32 unless otherwise noted. As
As did the SCSI interface adaptor 26, the NFS interface adaptor 28 includes a cache manager function 84 for managing accesses to the cache memory 42 (
A cache manager function 106 of the drive interface adaptor 46 manages data accesses with the cache memory 42. A Read/Write control function 104 handles the actual data flow, pursuant to a read or a write request, between the drive interface adaptor 46 and the associated physical disk unit 20.
Operation of the system of
If, on the other hand, the I/O request received from the host system 12 is a write request, the logical volume access function 64 will send the data block received with the request to the cache memory 42. Then, the logical volume access function 64 will, through the DIA interface function 66, send a write request to appropriate the drive interface adaptor 46, identifying the location in the cache memory 42 at which the data block to be written resides. The drive interface 46 will then access the cache memory 42 for the data block, and write it to physical storage 20.
File system requests are received by one of the file system interfaces: either the NFS, the CIFS, or the HTTP interface adapter, depending upon whether the source is a NFS, CIFS, or HTTP interface of the host system 12 and, therefore, one of the three protocols file system protocols: that is, NFS, CIFS, or HTTP. File system I/O requests may be accompanied by lock/unlock requests. A lock request seeks access to a specific data block within a specific file, or the file itself. An unlock request releases access to the block/file previously obtained. As is conventional, an lock or unlock request will include either the file name of the file sought to be accessed, or a block number in the specified file, and a block length. Alternatively, the request may include a file name and additional information identifying the right to access the file.
Control information for lock/unlock processing is stored in the cache memory 42 for the each of the protocols used by the file system interface adaptors 28, 30, 32, although other shared memory can be used if available.
File system I/O requests issued by the host system 12 are received by the TCP/IP interface function of the file system interface adaptor to which the request is directed. (That is, if an NFS host interface issues the request, the request will be received by the NFS interface adaptor 28. Similarly, for CIFS or HTTP host interfaces, the requests will be received by the CIFS or HTTP interface adaptors 30, 32 respectively. The requests will all, thereafter be handled in basically the same way as described hereinafter.) The TCP/IP interface function 70 will receive the request and pass it to the appropriate process function block 72 for further processing.
The process function block 72 will convert the received request to one for a common file system, and pass the converted request to the common file system function block 73 where it is received by a command process function 74 and transferred to a logical volume address converter function 76.
If the request is a lock request, it will also be passed to the lock manager function 78, which checks to determine whether or not access to the requested file is available. If access is available, the lock manager function 78 will initiate a reply (“access granted”) to the process function block 72. The process function block 72 will then notify the host system 12 of the access grant via the TCP/IP interface function 70. Generally, the locking protocol is specified in NFS, CIFS, or HTTP level. If, on the other hand, access is not available, for example being locked by another request, the lock manager function 78 will so notify the process function 72, which will send a request to host system 12 to pend the lock request. When the lock request is subsequently made available by release of the lock by the other request, the lock manager 78 will notify the host system 12 that access is granted.
I/O read or write requests from a file system interface of the host system 12 will include a file name, a block number in the specified file, and a block link. Read and write requests travel through the TCP/IP interface function 70, the process function block 72 and the command process function 74, to the logical volume address converter 76. There, the information in the request is converted to a logical volume unit number, a block number in the logical volume, and a logical block length. The logical address converter 76 will then pass this information to the logical volume access function block 64 which, as did the logical volume access function block 64 of the SCSI interface adaptor 26, will handle the data transfer in the same way; that is, if it is a read request, the logical volume access function block 82 will check to see if the requested information resides in the cache memory 42 and if so, retrieve the information and return it to the host system 12 in response to the request. If the requested information does not reside in the cache memory 42, the logical volume access function block 82 will issue a request to the appropriate drive interface adaptor 46, requesting that the information be retrieved from the physical storage 20. Write requests are also handled in the same manner as described above respecting the logical volume access block 64 of the SCSI interface adapter.
The drive interface adapters 46 will operate in the same manner when responding to read or write requests, regardless of the interface adapter issuing the request. It will execute read/write operations to and from the physical storage 20 in response to requests received from the interface adapters 26, . . . , 32. The drive interface adapters 46 preferably have the capability of performing write after processing from cache memory 42. (Write after processing is typically used, for example, in connection with mirrored storage. A write request will be processed by writing the data of the request to a specific physical storage unit 20. Subsequently, the same data, which may be stored in the cache memory 42, can be written to whatever disk storage unit (or units) 20 used for mirroring the data.)
The configuration of logical volumes may be established by a system administrator through a work station (not shown) connected to the storage controller 14 (
Logical volumes allow the physical storage 20 to be allocated between a block system and a file system as needed. For example, a first portion of the physical storage 20, say, one-third of the storage, may be allocated to block system data storage. Then, the remaining physical storage may be allocated to storing data for file system protocols. Later, it may be determined that less block system storage is actually needed so that the allocation could be changed, for example, something less than originally allocated, say one-fourth of the physical storage 20. The remaining physical storage 20 dedicated to file system storage is concomitantly increased.
Typically, logical volumes for a file system interface (e.g., the NFS or CIFS interface adapters 28, 30) will include file management information required by the common file system function block 73. This file management information provides the basis for the logical volume address conversion performed by the logical volume address converter 76 of the common file system block 73. Logical volume information for block system interface, i.e. the SCSI interface adapter 26, typically do not have such information, making it very difficult to access a logical volume for a block interface from a file interface. Therefore, in order to preclude unnecessary errors, status information can be included in each entry 122 for the logical volume, identifying whether that volume is a file system or a block system logical volume. Thus, as
There is, however, a way, according to the present invention, of accessing a logical volume for a file system from a block system interface, such as the SCSI interface adaptor 26. According to this aspect of the invention, that portion of the host system 12 having a SCSI interface is provided with a backup utility 12 a (
This will cause the entire logical volume identified in the request to be read from the physical storage 20, from the first address to the last address of the logical volume, without consideration of management information. The same portion of the host system 12 is also provided with the common file system library 12 b, which provides the ability to recognize the file management information of the common file system function 73. Thereby, the host system 12 can access an arbitrary file on a logical volume for a file system from an interface of a block system. (Thus, by using a common file system library, the host system 12 to access a file on a logical volume for a file interface through a block system interface (e.g., a SCSI interface, since a common file system library can recognize the file management information of the common file system function 73,)
In order to provide at least a modicum protection against inadvertent or other access of file system data from a block system interface or adapter, the logical volume table information could include information respecting whether or not the particular logical volume is accessible to certain types of access. For example, a file system logical volume would include information that it was or was not accessible from a block system access. Thus, as indicated in
Turning now to
Turning now to
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US5680537||Mar 1, 1995||Oct 21, 1997||Unisys Corporation||Method and apparatus for isolating an error within a computer system that transfers data via an interface device|
|US5828823||Apr 25, 1997||Oct 27, 1998||Unisys Corporation||Method and apparatus for storing computer data after a power failure|
|US5838950||Jun 7, 1995||Nov 17, 1998||Adaptec, Inc.||Method of operation of a host adapter integrated circuit|
|US5920893||Jun 2, 1997||Jul 6, 1999||Hitachi, Ltd.||Storage control and computer system using the same|
|US5983317||Feb 23, 1999||Nov 9, 1999||Hitachi, Ltd.||Storage device and method for data sharing|
|US6000020||Apr 1, 1997||Dec 7, 1999||Gadzoox Networks, Inc.||Hierarchical storage management from a mirrored file system on a storage network segmented by a bridge|
|US6115797||Aug 19, 1997||Sep 5, 2000||Hitachi, Ltd.||Method and system for sharing storage device via mutually different interfaces|
|US6145006||Jun 25, 1997||Nov 7, 2000||Emc Corporation||Method and apparatus for coordinating locking operations of heterogeneous host computers accessing a storage subsystem|
|US6195703||Jun 24, 1998||Feb 27, 2001||Emc Corporation||Dynamic routing for performance partitioning in a data processing network|
|US6230200||Sep 8, 1997||May 8, 2001||Emc Corporation||Dynamic modeling for resource allocation in a file server|
|US6233660||Nov 24, 1997||May 15, 2001||Emc Corporation||System and method for emulating mainframe channel programs by open systems computer systems|
|US6301605||Sep 29, 2000||Oct 9, 2001||Adaptec, Inc.||File array storage architecture having file system distributed across a data processing platform|
|US6304940||May 18, 1999||Oct 16, 2001||International Business Machines Corporation||Shared direct access storage system for MVS and FBA processors|
|US6324581||Mar 3, 1999||Nov 27, 2001||Emc Corporation||File server system using file system storage, data movers, and an exchange of meta data among data movers for file locking and direct access to shared file systems|
|US6425051||Sep 25, 2000||Jul 23, 2002||International Business Machines Corporation||Method, system, program, and data structures for enabling a controller accessing a storage device to handle requests to data in a first data format when the storage device includes data in a second data format|
|US6430660||May 21, 1999||Aug 6, 2002||International Business Machines Corporation||Unified memory hard disk drive system|
|US6446141||Mar 25, 1999||Sep 3, 2002||Dell Products, L.P.||Storage server system including ranking of data source|
|US6505273||Nov 24, 1998||Jan 7, 2003||Fujitsu Limited||Disk control device and method processing variable-block and fixed-block accesses from host devices|
|US6516351||Oct 21, 1998||Feb 4, 2003||Network Appliance, Inc.||Enforcing uniform file-locking for diverse file-locking protocols|
|US6549988||Jan 22, 1999||Apr 15, 2003||Ilya Gertner||Data storage system comprising a network of PCs and method using same|
|US6574667||Jan 2, 2001||Jun 3, 2003||Emc Corporation||Dynamic routing for performance partitioning in a data processing network|
|US6598129||Jun 5, 2001||Jul 22, 2003||Motohiro Kanda||Storage device and method for data sharing|
|US6606690||Feb 20, 2001||Aug 12, 2003||Hewlett-Packard Development Company, L.P.||System and method for accessing a storage area network as network attached storage|
|US20010037406||Mar 9, 2001||Nov 1, 2001||Philbrick Clive M.||Intelligent network storage interface system|
|US20020095547||Jan 12, 2001||Jul 18, 2002||Naoki Watanabe||Virtual volume storage|
|US20020156984||Feb 20, 2001||Oct 24, 2002||Storageapps Inc.||System and method for accessing a storage area network as network attached storage|
|US20020161855||Dec 5, 2000||Oct 31, 2002||Olaf Manczak||Symmetric shared file storage system|
|US20030046357||Aug 30, 2001||Mar 6, 2003||International Business Machines Corporation||Intelligent content placement in a distributed computing network|
|US20030110237||Sep 25, 2002||Jun 12, 2003||Hitachi, Ltd.||Methods of migrating data between storage apparatuses|
|US20030120743||Dec 21, 2001||Jun 26, 2003||Coatney Susan M.||System and method of implementing disk ownership in networked storage|
|US20030126523||Dec 28, 2001||Jul 3, 2003||Corbett Peter F.||Row-diagonal parity technique for enabling efficient recovery from double failures in a storage array|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US7644204||Oct 31, 2006||Jan 5, 2010||Hewlett-Packard Development Company, L.P.||SCSI I/O coordinator|
|US20080147893 *||Oct 31, 2006||Jun 19, 2008||Marripudi Gunneswara R||Scsi i/o coordinator|
|U.S. Classification||710/74, 710/33, 711/100, 709/233, 710/2|
|International Classification||G06F13/10, G06F13/12, G06F3/06, G06F12/00, G06F13/14|
|Cooperative Classification||Y10S707/99931, G06F3/0689, G06F3/0661, G06F3/0617, G06F3/0643, G06F3/0626, G06F3/0665, G06F17/30197, G06F3/064, G06F3/061, G06F3/0619, G06F3/067, G06F3/0605, G06F11/2056, G06F3/065, H04L67/1095, H04L67/1097|
|European Classification||G06F3/06A4F2, G06F3/06A4F4, G06F3/06A6L4R, G06F3/06A2R4, G06F3/06A2P, G06F3/06A4H4, G06F3/06A6D, G06F3/06A2A2, G06F3/06A2R6, G06F17/30F8D1, G06F3/06A4T8, G06F3/06A4V4|
|Aug 24, 2009||FPAY||Fee payment|
Year of fee payment: 4
|Aug 21, 2013||FPAY||Fee payment|
Year of fee payment: 8